NEW ALGORITHM FOR UNSUPERVISED CLASSIFICATION

This Letter presents a new algorithm for the unsupervised classification of multispectral remotely-sensed image data. A user inputs the desired number of classes without the introduction of their initial class centres. The algorithm automatically creates the class centres according to the separability of data. The performance of the approach has been evaluated through an example of unsupervised classification of Thematic Mapper multispectral image data.